| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
|
FJ
EXPRESS SUMMARY ARTICLE The Full-length version of this article is also available, published online December 28, 2001 as doi:10.1096/fj.01-0343fje. |
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Institute for Biological Sciences, National Research Council of Canada, Ottawa, Canada K2B 8G2
2Correspondence: Institute for Biological Sciences, National Research Council of Canada, Montreal Road Campus, Bldg. M-54, Ottawa, ON, Canada, K1A 0R6. E-mail: danica.stanimirovic{at}nrc.ca
SPECIFIC AIMS
The aim of these studies was to discover novel receptor–ligand interactions at the blood–brain barrier (BBB) essential for developing brain targeting molecules with improved vectoring capabilities. A more specific goal was to design and apply an effective approach to select for and characterize novel BBB binding and transmigrating antibodies using an antibody phage display technology.
PRINCIPAL FINDINGS
1. Two novel BBB-specific, phage-displayed single-domain antibodies (sdAbs) were selected from llama sdAb phage display library using an original subtractive panning and functional selection protocol shown in Fig. 1
.
A nonimmunized llama sdAb phage display library (5.6x108 species) derived from the VHH of the heavy chain immunoglobulin Gs (IgGs) that occur naturally in the absence of light chain was used for panning. The average molecular mass of these sdAbs is 14 kDa, approximately half the size of an scFv and 10-fold smaller than IgG molecule.
|
The nonimmunized phage library was preabsorbed onto human lung microvascular endothelial cells (HLMEC) to remove common endothelial binding VHHs and then applied to human cerebromicrovascular endothelial (HCEC) target cells (Fig. 1A
). The bound phage was dislodged by three rounds of highly stringent stripping washes (S1, S2, and S3); HCEC were then lysed to capture the internalized phage (Int). Four rounds of panning were performed; for each round, the input phage was derived from the output fraction showing the highest percent of phage with the full-length VHH. After four rounds of panning, phage sdAbs were screened for the 1) specificity for HCEC in a phage ELISA (Fig. 1B
, left arm) and 2) ability to transmigrate across in vitro BBB model (Fig. 1B
, right arm).
Of 58 clones shown to have full-length VHH by PCR, 16 were specific for HCEC vs. HLMEC or human umbilical vein endothelial cells (HUVEC) in phage ELISA. These clones belonged to two different sequences, designated FC5 (GenBank acc. AF441486) and FC44 (GenBank acc. AF441487) (Fig. 1C
).
Phage-displayed sdAbs were also screened for their ability to cross an in vitro BBB model (Fig. 1B
). The human in vitro BBB model used for these studies consists of a monolayer of primary HCEC grown on a porous membrane positioned between two media compartments, as shown in Fig. 2
A. The integrity and tightness of HCEC monolayers were monitored by measuring diffusion of the paracellular marker sodium fluorescein [permeability coefficient (3.2±0.3) x 10-3 cm/min]. HCEC monolayers were impermeable for 10 and 70 kDa dextran. 1011 pfu phage amplified from the fourth round S3 and Int fractions were applied to the top chamber of the in vitro BBB model, and titers in the bottom chamber were determined at time intervals. Both the wild-type phage and the recombinant phage displaying an unrelated VHH (NC11) produced low (120–200 pfu) phage titers in the bottom chamber 60–90 min after addition to the top compartment (Fig. 2B
). In contrast, the S3 and Int phage produced high titers in the bottom chamber as early as 15 min after addition, with titers reaching 5000–7000 pfu at 60–90 min (Fig. 2B
). Empty membrane did not restrict the passage of phage. Sequencing of phage clones that transmigrated in vitro BBB model revealed that all phage clones with the full-length inserts were either FC5 or FC44 (Fig. 2C
). Plaque PCR and sequencing of full-length clones applied to the top chamber showed clones with various other sequences.
|
2. FC5- and FC44-c-myc-His5 fusion proteins were internalized by live HCEC and transmigrated across an in vitro BBB model
To assess sdAb binding and uptake by HCEC, FC5 and FC44, as well as a negative control sdAb (NC11) were expressed in Escherichia coli as fusion proteins with c-myc and His5 tags and purified. Binding of sdAbs to cell lysates of endothelial cells from the brain (HCEC and human pial artery endothelial cells) or peripheral organs (HLMEC and HUVEC), fetal human astrocytes (FHAs), or live HCEC and HLMEC was assessed by ELISA against c-myc. FC5 and FC44 demonstrated highly selective binding to endothelial cells derived from the brain vasculature. FC5 and FC44 labeled with the fluorescent dye Alexafluor 488 were taken up by live HCEC but not HLMEC.
FC5 and FC44 are also shown to transmigrate across the human in vitro BBB model with clearance values comparable to those seen across empty membranes (2.28±0.27 vs. 2.42±0.15 µl/min for FC5; 3.23±0.62 vs. 2.51±0.21 µl/min for FC44, respectively) whereas HCEC were impermeable for NC11 (0.097±0.005 vs. 2.33±0.27 µl/min, respectively). The concentration of sdAbs in the bottom chamber was determined by immobilizing sdAbs on nickel-coated wells and performing ELISA against c-myc.
3. FC5 and FC44 were detected in brain tissue after intravenous (i.v.) injection into mice
The ability of FC5 and FC44 to target the brain in vivo was assessed by injecting phage-displayed (109 pfu) or soluble (30 µg/mice) FC5 and FC44 into mice i.v. Brain uptake of phage-displayed FC44 and FC5 (Fig. 3
A) was 4.5 ± 2.7% and 2.9 ± 1.7% of injected dose/gram tissue, respectively, in contrast to negligible brain uptakes of phage-displayed NC11 and wild-type phage (<0.1%). Uptake of phage-displayed FC5 and FC44 by lung tissue was significantly lower than that of NC11 and wild-type phage (Fig. 3A
). Kidney and liver, likely elimination routes, contained high phage titers of FC44, FC5, NC11, and wild-type phage (Fig. 3A
).
|
FC5, FC44, and NC11 fusion proteins were extracted from mouse tissues by affinity purification and were detected on a Western blot using an anti-c-myc antibody (Fig. 3B
). Both FC5 and FC44, but not NC11, were detected in the brain extract after microvessel depletion (Fig. 3B
). Kidney tissue showed strong bands for all three sdAb fusion proteins (Fig. 3B
).
CONCLUSIONS AND SIGNIFICANCE
We developed an original selection approach to extricate BBB-selective binders from the phage display library of llama sdAbs by 1) subtractive panning of the library against human peripheral (i.e., lung) and brain endothelial cells; 2) selecting phage clones that selectively bind and are internalized by human brain endothelial cells; and 3) selecting phage clones that transmigrate across human in vitro BBB model. This strategy yielded two candidate antibodies, FC5 and FC44, which were then shown to target the brain in vivo in a partly selective manner. It is also shown that small peptides (Mr of c-myc-His5, tag 
2.5 kDa; 26 amino acids) can be attached to and transported together with FC5 and FC44 as fusion proteins across the BBB in vitro and in vivo. Therefore, vectoring properties of these sdAbs can be exploited to develop efficient antibody carriers suitable for delivery of macromolecules across the human BBB.
The delivery of therapeutics across the BBB remains one of the most perplexing challenges in developing treatments for neurological diseases. Current brain drug delivery practices use invasive neurosurgical procedures or pharmacological methods to improve drug penetration into the brain. Macromolecule delivery across the BBB using antibody ‘vectors’ takes advantage of transporters or receptors selectively expressed on brain capillary endothelium, such as transferrin receptor, that undergo receptor-mediated transcytosis. As only few such receptors are currently known, discovery of novel receptor–ligand interactions at the BBB is essential for developing brain targeting molecules with better vectoring capabilities.
FC5 and FC44 may have several advantages as antibody vectors over currently used/tested antibodies. Although detailed pharmacokinetic studies are needed to determine whether small size of sdAbs is advantageous for better tissue penetration, the nonspecific interaction of sdAbs with tissues expressing high levels of Fc receptors (e.g., liver, spleen) is expected to be low, since sdAbs lack Fc domain. Moreover, sdAbs are shown to have a remarkable stability against high temperature, pH, and salts and are typically expressed in high quantity in the E. coli periplasm.
The identification and characterization of antigen epitopes recognized by FC5 and FC44 will allow for engineering of sdAbs to improve vectoring properties. FC5 and FC44 sequences can then be used as templates for molecular modeling of small drugs/ligands against their respective antigens. Basic principles of the approach used in this study are applicable to other research fields including cancer immunotherapy, gene therapy of tumors and vascular diseases, and drug delivery in general.
FOOTNOTES
1 To read the full text of this article, go to http://www.fasebj.org/cgi/doi/10.1096/fj.01-0343fje; to cite this article, use FASEB J. (December 28, 2001) 10.1096/fj.01-0343fje 
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
